Long post but may be useful to others wanting to design their own speakers.
OK,
So off I toddled to the MCRU speaker bakeoff with the semi-omnis and as soon as I saw the room, superb that it is, I thought I'd better get my coat. My farty little semi-omnis were dwarfed by everything else in the room. And I left them in the little vestibule, just outside the room so I could sneak them back into the car before anyone noticed.
However after the Doc and Savvy had demmed a pair of cubes I felt a bit better.
The OmniMets performed better than I expected, given the size of the room and produced a sound devoid of boom and with plenty of rhythm, tunes and drive. The bass end worked very well and the near wall placement incorporated into the design of the things was spot on. Those were the positives.
HOWEVER all was far from well at the upper mid and lower treble boundary. Out of the bass and lower mids area they sounded as flat as a fart, swathes of detail and harmonics missing in action, boring as hell and at least 70% of the people in the room just ignored them and started having conversations.
That absolutely had not happened with any of the other speakers in the room and frankly I was fucking pissed off, big style, quite angry in fact and it was written all over my face how I was feeling. Poker faced I am not unfortunately, which is not good in these sorts of "public" circumstances because it inevitably comes over as "sore loser" syndrome.
After leaving the room, taking a walk to Tescos and calming down I pondered what had happened and put on my objective head. Some of the audience had lost interest for a good reason, they didn't do it on purpose just to piss me
me off, and a couple of the guys there struck up a conversation about harmonic details of guitar strings, cymbals being shut off because there was a hole in the treble response, air and space just not happening, dry as dust.
PRaT was excellent, bass even and punchy, not a lot else but they played alright. Jesus! I began to realise that I had actually gone and designed a bloody Flat Earth speaker and after all I had said about never going there again, there I was once more, and to make matters worse, I had done it without any help from Linn/Naim
This is not something I had noticed at home and neither had anyone else who had heard them in my own room, in fact they had been used in several rooms and the problem had never reared its head, but it bloody well had now! Indeed they had measured relatively flat through the third octave spectrum analyser, the measurements appearing to confirm their goodness.
BUT this had by now happened twice; once at Owston and now in MCRU's dem room, so something was going pear shaped when they were taken out of their domestic context, which I'm afraid is just not good enough in my book. Just goes to show that measurements need to be treated with a healthy degree of skepticism and caution.
OK so between us, myself Colin and Chris, after a long conversation out of the dem room had agreed that it had to be the crossover to the tweeter that was causing the problem with the missing bits of the music and I resolved to do something about it as soon as I got home.
Looking at my calculations for the crossover capacitor value, it all looked right and double checking it all, it was still correct, so no clue as to what was going on. I examined the 45 degree, off-axis response curve of the bass mid driver, with its white tack induced roll off and the curve that the first-order tweeter crossover capacitor gave with my calculation, and in theory, the curves married up to give a flat response. In other words, on paper, everything lined up perfectly. There was much head scratching and beard ruffling (well there would have been, if I'd had a beard) I mean, what the hell was going on?
This was getting a bit silly now. I've been building speakers for nigh on forty years. OK, yes, granted; this is a semi-omni, but come on, I do know what I'm doing. I'm missing something obvious, but I'm damned if I can work out what it is, these things should work and they do.....yet they don't if you get my drift.
As a last resort I went right back to the SB Acoustics catalogue and got out the spec for the tweeter:
SB.Acoustics SB29RDNC-C000-4 Tweeter
Efficiency, 94dB
Impedance - 8 ohms?.....err...no...4 Ohms.....OH FOR FUCK'S SAKE
The carefully chosen crossover cap was half the value it should have been, meaning the crossover frequency was twice as high as it should have been, hence the hole in the treble. Oh poo!
So the cap value was doubled with another one the same value in parallel with it and it all fell into place soundwise, musicwise and everything elsewise. The other three omnis I have built all had the same error in the crossover, due to me somehow misreading the HF driver data.
Here endeth the first lesson. Check, check and check again. Make sure you have the bloody right data before you design your crossover!